Evaluating the contribution of the CHAMELEON harmonisation assistant in human musical creativity (MIS 5005182)
Asterios Zacharakis, Maximos Kaliakatsos-Papakostas, Stamatia Kalaitzidou and Emilios Cambouropoulos

Summary

This research project implemented a scientific protocol for evaluating the effect of the CHAMELEON melodic harmonisation assistant on human musical creativity. CHAMELEON's innovation lies in the fact that, by applying principles of the cognitive theory of Conceptual Blending, it allows the blending of harmonic characteristics of different musical idioms with the aim of inventing new harmonic 'hybrids'. Thus, it can provide a variety of different —and often novel— melodic harmonisations of a given melodic material. In order to investigate the potential creative influence of CHAMELEON on human users, an online interface was created which enables the use of the computer system by following some simple instructions. A specially designed behavioural experiment evaluated CHAMELEON's contribution to the creative process of melodic harmonisation by comparing metrics of user experience as well as characteristics of the harmonisations per se, created with and without the use of CHAMELEON.

Main findings

Two types of data were collected from the evaluation experiment. The first type involved behavioural data assessing user experience for creating melodic harmonisation of a given melody with and without computational assistance. The participants of the experiment were divided into two groups based on their experience in composition: professional composers and musicology students. The analysis of user experience showed that in general the computational tool was appreciated more by students than by professional composers. More specifically, the students stated that the use of CHAMELEON helped them to express their ideas better even though the process seemed more cognitivelly demanding than the harmonisation without computational support. In addition, the group of students scored higher than the professionals in certain metrics suggesting that the use of CHAMELEON enabled them to become more creative and that it provided them with some good ideas that they would have not easily reached on their own.

The second type of data concerned the computational analysis of the harmonisations submitted by the participants at the end of the two experimental procedures (simple and computationally supported). This type of analysis of the final products of the use of CHAMELEON showed that the computer-assisted harmonisations were on average more complex, more varied and more unexpected than unassisted harmonisations. Finally, the qualitative analysis of each participant's harmonisations in comparison to the favoured harmonisations selected from CHAMELEON's examples by each participant, showed that the majority actually used information from these harmonisations either in the form of copying single chords or exact chord sequences, or (less frequently) by adopting certain abstract concepts.

In conclusion, the research results of this project suggest that CHAMELEON seems to meet the basic requirements for a computational support system in the field of music creativity.